| TheJulia | he wins the vacation contest | 00:52 |
|---|---|---|
| opendevreview | Merged openstack/ironic master: Consolidate duplicated TLS code into ironic/common/tls_utils https://review.opendev.org/c/openstack/ironic/+/987909 | 03:56 |
| opendevreview | Riccardo Pittau proposed openstack/ironic master: Fix kernel parameter parsing for quoted values and whitespace https://review.opendev.org/c/openstack/ironic/+/992025 | 07:16 |
| rpittau | this ^ needs some attention :) | 07:17 |
| *** elodilles_OoO is now known as elodilles | 08:36 | |
| opendevreview | Konstantinos Karampogias proposed openstack/ironic master: Add vendor/model/serial_number to firmware component endpoint https://review.opendev.org/c/openstack/ironic/+/992034 | 08:41 |
| dtantsur | rpittau: I'm really curious why both BMO and Bifrost are happy on https://review.opendev.org/c/openstack/ironic/+/991365 | 09:35 |
| dtantsur | where did you notice the failure? | 09:35 |
| opendevreview | Konstantinos Karampogias proposed openstack/ironic master: Add vendor/model/serial_number to firmware component endpoint https://review.opendev.org/c/openstack/ironic/+/992034 | 09:40 |
| opendevreview | Dmitry Tantsur proposed openstack/ironic master: DNM check standalone jobs with broken inspection https://review.opendev.org/c/openstack/ironic/+/992202 | 09:45 |
| janders | TheJulia https://bugs.launchpad.net/ironic/+bug/2154614 done | 10:45 |
| opendevreview | OpenStack Release Bot proposed openstack/ironic-python-agent bugfix/11.6: Update .gitreview for bugfix/11.6 https://review.opendev.org/c/openstack/ironic-python-agent/+/992207 | 10:46 |
| opendevreview | Jacob Anders proposed openstack/ironic-specs master: Add spec for multi-component batched firmware updates https://review.opendev.org/c/openstack/ironic-specs/+/992208 | 10:46 |
| opendevreview | Jacob Anders proposed openstack/ironic-specs master: Add spec for multi-component batched firmware updates https://review.opendev.org/c/openstack/ironic-specs/+/992208 | 10:47 |
| rpittau | dtantsur: I've noticed the failure downstream | 11:15 |
| dtantsur | rpittau: puzzling. I don't think we even use kernel parameters downstream.. | 11:21 |
| dtantsur | We definitely cover inspection in standalone jobs as evident by https://review.opendev.org/c/openstack/ironic/+/992202 | 11:22 |
| opendevreview | Jacob Anders proposed openstack/ironic-specs master: Add spec for multi-component batched firmware updates https://review.opendev.org/c/openstack/ironic-specs/+/992208 | 11:39 |
| opendevreview | Jacob Anders proposed openstack/ironic-specs master: Add spec for multi-component batched firmware updates https://review.opendev.org/c/openstack/ironic-specs/+/992208 | 11:41 |
| opendevreview | Jacob Anders proposed openstack/ironic-specs master: Add spec for multi-component batched firmware updates https://review.opendev.org/c/openstack/ironic-specs/+/992208 | 11:42 |
| hjensas | Last PTG we talked about gathering all "touch network devices" code in NGS. I proposed a series of changes to migrate the ML2 Netconf drive in networking-baremtal that need reviews. https://review.opendev.org/q/topic:%22netconf%22 | 12:07 |
| opendevreview | cid proposed openstack/ironic master: Stop skipping Redfish vmedia insert on stale BMC state https://review.opendev.org/c/openstack/ironic/+/990925 | 12:22 |
| TheJulia | good morning | 13:12 |
| opendevreview | Konstantinos Karampogias proposed openstack/ironic master: Add vendor/model/serial_number to firmware component endpoint https://review.opendev.org/c/openstack/ironic/+/992034 | 13:19 |
| opendevreview | Harald JensĂĄs proposed openstack/networking-generic-switch master: Pass physnet VLAN segments to device drivers https://review.opendev.org/c/openstack/networking-generic-switch/+/990510 | 13:38 |
| dtantsur | I'm looking at BMO job logs and thinking that I"m probably going fully insane (finally): https://paste.opendev.org/show/bsrbdhKMUyzzbjJ9q6PZ/ | 14:18 |
| dtantsur | TheJulia: any idea how references to "metalsmith" could get there ^^? O_o | 14:18 |
| * TheJulia blinks | 14:19 | |
| TheJulia | no idea, base image starting point maybe? | 14:20 |
| dtantsur | TheJulia: the image is built by Ironic... | 14:20 |
| TheJulia | vmedia attachment with sushy-tools? | 14:21 |
| dtantsur | TheJulia: yep | 14:22 |
| dtantsur | This is *without* actual network data provided. I wonder if Ironic can set a non-sensual value in this case.. | 14:23 |
| TheJulia | pre-baked ISO ? | 14:23 |
| dtantsur | TheJulia: to my best knowledge, no.. I'll double-check | 14:24 |
| TheJulia | Yeah, I think we retain and carry across values | 14:24 |
| dtantsur | TheJulia: the problem is: we also set ir_pub_id, so the start-up job loops 30 seconds before bailing and letting IPA continue | 14:25 |
| dtantsur | it's half a minute for each test run | 14:25 |
| TheJulia | yeah, seems super weird because I have no idea where that values is coming from | 14:26 |
| dtantsur | I don't see any reference to metalsmith anywhere. The command line seems correct. | 14:26 |
| * dtantsur is finally insane, yay | 14:26 | |
| dtantsur | In the ironic logs, I see a reasonable publisher provided | 14:27 |
| dtantsur | Ahhh, wait, it's a configdrive picked up from the disk. Cirros? | 14:29 |
| dtantsur | TheJulia: shouldn't we bail immediately on publisher_id that is not empty and does not match the expected one? | 14:29 |
| TheJulia | dtantsur: "finallly"? | 14:30 |
| TheJulia | dtantsur: oh yeah, since it looks like a disk I guess | 14:30 |
| TheJulia | dtantsur: likely we should do that | 14:30 |
| * dtantsur files a bug | 14:30 | |
| dtantsur | It was quite intentional in https://opendev.org/openstack/ironic-python-agent-builder/commit/96d997f15f9f8a3c8d155a665a2ce0972d34cc61 though | 14:33 |
| dtantsur | Right, yeah, their reasoning is not terrible | 14:33 |
| dtantsur | TheJulia: maybe it's the opposite situation: we should not create a publisher ID if there is no configuration to inject? | 14:35 |
| * TheJulia shrugs | 14:35 | |
| TheJulia | Alternatively, we should validate more aggressively that the volume *IS* a vmedia ISO I guess and not a local disk. | 14:36 |
| TheJulia | removing the publisher ID prevents us from blocking/locking out afaik | 14:36 |
| dtantsur | I'd not bet on all virtual media implementation doing the right thing.. | 14:36 |
| dtantsur | right | 14:36 |
| TheJulia | yeah, I think my concern is if we special case it out, then we re-create the same issue we were seeking to avoid originally | 14:36 |
| TheJulia | or at least, begin heading back in that direction | 14:37 |
| * dtantsur thinks again | 14:37 | |
| dtantsur | Hold on, it does not make any sense. If we create a publisher ID and pass it through the kernel params, why doen's the script find the device? | 14:37 |
| * dtantsur thinks more | 14:37 | |
| dtantsur | Probably because we create a publisher ID but do not use the label config-2 | 14:38 |
| TheJulia | I think we do use the config-2 label, but you risk a local partition already having the label | 14:40 |
| TheJulia | the publisher_id is the only way we have in the code path to identify and match up volume to boot | 14:40 |
| dtantsur | Maybe use config-2 unconditionally? | 14:40 |
| dtantsur | Anyway, I've filed https://bugs.launchpad.net/ironic-python-agent-builder/+bug/2155910. I'm surely not going to find time for this too.. | 14:41 |
| TheJulia | oh yeah, so we should disqualify a device which is not labeled | 14:41 |
| TheJulia | which would prevent the issue youre encountering | 14:41 |
| dtantsur | the problem is that the local partition is labeled and the ISO is not | 14:42 |
| TheJulia | then there was no data injected? | 14:42 |
| dtantsur | TheJulia: no data and nothing is really broken, except that we're waiting for publisher ID that is never going to appear | 14:44 |
| TheJulia | Yeah, we should likely create a list of "I've seen this already, and its not it" | 14:45 |
| TheJulia | and wait out the loop otherwise, but you still have the pause | 14:45 |
| dtantsur | Oh! I think we need a short-cut in case where a device with the publisher ID exists but does not have label config-2 | 14:45 |
| TheJulia | ding ding ding | 14:46 |
| opendevreview | Michal Nasiadka proposed openstack/bifrost master: [WIP] Experimental support for ARM64 hosts on Debian https://review.opendev.org/c/openstack/bifrost/+/920057 | 14:48 |
| TheJulia | Everyone getting caffinated? | 14:56 |
| JayF | I just sat down 5 minutes ago with a cuppa joe and donut from the local donut shop | 14:58 |
| TheJulia | yum | 14:58 |
| JayF | however I am NOT sufficiently caffinated, because their coffee machine temp is set to "surface of sun" | 14:58 |
| TheJulia | Is that more or less than "Mcdonalds scald" ? | 14:59 |
| JayF | it's so hot it hurts to hold even with the little coffee jacket installed | 15:00 |
| JayF | #startmeeting ironic | 15:00 |
| opendevmeet | Meeting started Mon Jun 8 15:00:09 2026 UTC and is due to finish in 60 minutes. The chair is JayF. Information about MeetBot at http://wiki.debian.org/MeetBot. | 15:00 |
| opendevmeet | Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. | 15:00 |
| opendevmeet | The meeting name has been set to 'ironic' | 15:00 |
| JayF | Hey all, welcome to the Ironic meeting. We're following the OpenInfra Code of Conduct. | 15:00 |
| JayF | Agenda is at https://wiki.openstack.org/wiki/Meetings/Ironic | 15:00 |
| JayF | #topic Announcements/Reminders | 15:00 |
| JayF | Please review things hashtag:ironic-week-prio. Need reviews? Add that hashtag. | 15:00 |
| JayF | #link https://tinyurl.com/ironic-weekly-prio-dash | 15:00 |
| kubajj | o/ | 15:01 |
| JayF | I'll give us a few minutes for quorum to develop | 15:01 |
| rpittau | o/ | 15:01 |
| TheJulia | o/ | 15:01 |
| Mahnoor | o/ | 15:01 |
| * TheJulia wonders if a "qourum compiler" exists | 15:01 | |
| dtantsur | o/ | 15:01 |
| mumesan[m] | o/ | 15:03 |
| cid | o/ | 15:04 |
| JayF | Alright! First up | 15:05 |
| JayF | #topic Working Group Updates | 15:05 |
| JayF | Anything to speak to here around Async IO? | 15:05 |
| JayF | That's a lot of silence I'm going to take as a no. | 15:06 |
| JayF | The other working group item I added here was around Security | 15:06 |
| JayF | for some context; there's an Ironic "coresec" team -- myself, TheJulia, rpittau, dtantsur -- we are the folks who initially get access to security sensitive bug reports and deal with their disposition. | 15:07 |
| JayF | And we've been busy! | 15:07 |
| TheJulia | exceptionally so. | 15:07 |
| JayF | Last week, Ironic released three OSSA (advisories) and one OSSN (note). | 15:07 |
| JayF | #link https://security.openstack.org/ossa/OSSA-2026 | 15:08 |
| JayF | #undo | 15:08 |
| opendevmeet | Removing item from minutes: #link https://security.openstack.org/ossa/OSSA-2026 | 15:08 |
| JayF | #link https://security.openstack.org/ossa/OSSA-2026-017.html | 15:08 |
| JayF | #link https://security.openstack.org/ossa/OSSA-2026-018.html | 15:08 |
| JayF | #link https://security.openstack.org/ossa/OSSA-2026-019.html | 15:08 |
| JayF | #link https://wiki.openstack.org/wiki/OSSN/OSSN-0099 | 15:09 |
| JayF | Please let someone know if you have any issue with the fixes. | 15:10 |
| JayF | #topic Discussion Topics | 15:10 |
| JayF | Improvement to firmware updates -- PTG followup has iurygregory (not here?) and janders listed | 15:10 |
| dtantsur | I think the conversation happened last time | 15:11 |
| Mahnoor | I *think* this was done last meeting, yes | 15:11 |
| JayF | ack; moving on | 15:11 |
| JayF | #topic Bug Deputy Updates | 15:11 |
| TheJulia | Sort of did, jacob followed up with an initial spec super early this morning | 15:11 |
| JayF | Mahnoor is listed as bug deputy | 15:11 |
| JayF | #undo | 15:12 |
| opendevmeet | Removing item from minutes: #topic Bug Deputy Updates | 15:12 |
| Mahnoor | 3 new bugs in ironic, and 1 RFE | 15:12 |
| JayF | TheJulia: the spec isn't attached to the bug in the agenda | 15:12 |
| JayF | Mahnoor: hang on, wanna make sure janders' work gets recognized properly | 15:12 |
| Mahnoor | ack | 15:12 |
| JayF | #link https://review.opendev.org/c/openstack/ironic-specs/+/992208 | 15:12 |
| JayF | Please take time and review the spec posted by janders this morning :) | 15:13 |
| JayF | now to kick it over to Mahnoor :D | 15:13 |
| JayF | #topic Bug Deputy Updates | 15:13 |
| Mahnoor | There were 3 new bugs in ironic :) | 15:13 |
| Mahnoor | I've mentioned in the notes the links | 15:13 |
| JayF | Do we have a volunteer to bug deputy next week? | 15:14 |
| cid | I don't mind. | 15:15 |
| JayF | #action cid is bug deputy next week | 15:15 |
| cid | By the way, kudos to the coresec team on security fixes so far | 15:15 |
| Mahnoor | ++, must be a lot | 15:16 |
| cid | Yup :-) | 15:16 |
| JayF | #topic RFE Review | 15:16 |
| JayF | We have one listed in here, hopefully not stale | 15:16 |
| JayF | #link https://bugs.launchpad.net/ironic/+bug/2154817 | 15:17 |
| JayF | it was commented that it was a duplicate of | 15:17 |
| JayF | #link https://bugs.launchpad.net/ironic/+bug/2098693 | 15:17 |
| Mahnoor | Its part of Outreachy, it says. But I was a little confused by multiple people working on it..and why there was a need to duplicate | 15:18 |
| Mahnoor | Its not a crisis though :) | 15:18 |
| TheJulia | i, too, have been confused about this | 15:18 |
| JayF | a lot of the things being worked on are things we'd often ask for a spec for, too | 15:19 |
| JayF | like API changes | 15:19 |
| rpittau | me too, and I'm directly involved :D | 15:19 |
| JayF | it seemed a little weird to have a bunch of new folks drop in working on an API change we didn't have specified | 15:19 |
| rpittau | that was a misunderstanding | 15:19 |
| TheJulia | rpittau: even better! | 15:19 |
| JayF | I know we are often loosey goosey with specs so I didn't wanna assume lol | 15:19 |
| TheJulia | I feel like, yes, there is a validity to what multiple folks seem to be trying to drive forward | 15:20 |
| TheJulia | There are some baseline challenges though, I want them all to be successful | 15:20 |
| TheJulia | obviously, that is not *always* possible | 15:21 |
| rpittau | we've reduced the scope of https://review.opendev.org/c/openstack/ironic/+/983349 | 15:21 |
| rpittau | our outreachy intern estherd[m] will work on the rest | 15:21 |
| TheJulia | is this just a case of others seeing the project and maybe thinking they could possibly still qualify if they go work it separately? | 15:21 |
| rpittau | no, it was mistaken as something open to all the outreachy candidates | 15:22 |
| Mahnoor | or they thought it was part of the contribution phase? 🤔 | 15:22 |
| rpittau | yeah | 15:22 |
| TheJulia | oh, i see | 15:22 |
| rpittau | kind of | 15:22 |
| JayF | So we should mark that new bug as a dupe? Or the old one? | 15:24 |
| JayF | What's the adjudication here for Mahnoor to execute | 15:24 |
| rpittau | not a duplicate, we'll close the fdirst one once the relataed patch merges | 15:24 |
| JayF | ack, okie dokie :D | 15:25 |
| rpittau | thanks :) | 15:25 |
| JayF | #topic Open Discussion | 15:25 |
| JayF | Anything to discuss not on the agenda? | 15:25 |
| TheJulia | i've got nothing right now | 15:26 |
| TheJulia | oh! | 15:26 |
| TheJulia | Last call for: https://review.opendev.org/c/openstack/ironic-specs/+/954612 | 15:26 |
| TheJulia | And we can close out the meeting then | 15:28 |
| JayF | nope, one more | 15:28 |
| JayF | Who is chairing next week? | 15:28 |
| JayF | Do we have a volunteer? | 15:28 |
| TheJulia | I can | 15:29 |
| JayF | sgtm, | 15:29 |
| JayF | #action TheJulia to emcee the next meeting | 15:29 |
| JayF | Thanks all, meeting over o/ | 15:29 |
| JayF | #endmeeting | 15:29 |
| opendevmeet | Meeting ended Mon Jun 8 15:29:19 2026 UTC. Information about MeetBot at http://wiki.debian.org/MeetBot . (v 0.1.4) | 15:29 |
| opendevmeet | Minutes: https://meetings.opendev.org/meetings/ironic/2026/ironic.2026-06-08-15.00.html | 15:29 |
| opendevmeet | Minutes (text): https://meetings.opendev.org/meetings/ironic/2026/ironic.2026-06-08-15.00.txt | 15:29 |
| opendevmeet | Log: https://meetings.opendev.org/meetings/ironic/2026/ironic.2026-06-08-15.00.log.html | 15:29 |
| rpittau | thanks | 15:29 |
| opendevreview | Merged openstack/ironic-specs master: Deferred Actions https://review.opendev.org/c/openstack/ironic-specs/+/954612 | 16:10 |
| opendevreview | Julia Kreger proposed openstack/ironic master: Security: Fix sensitive properties returned on volume targets https://review.opendev.org/c/openstack/ironic/+/990430 | 16:30 |
| opendevreview | Clif Houck proposed openstack/ironic master: Add a custom configuration type KernelParameterString https://review.opendev.org/c/openstack/ironic/+/992306 | 16:31 |
| opendevreview | Clif Houck proposed openstack/ironic master: Add kernel parameter parsing to create_boot_iso https://review.opendev.org/c/openstack/ironic/+/992307 | 16:31 |
| opendevreview | Clif Houck proposed openstack/ironic master: Add kernel parameter parsing to build_extra_pxe_options https://review.opendev.org/c/openstack/ironic/+/992308 | 16:31 |
| opendevreview | Clif Houck proposed openstack/ironic master: Rip out and replace ironic.common.utils.parse_kernel_params https://review.opendev.org/c/openstack/ironic/+/992309 | 16:31 |
| opendevreview | Julia Kreger proposed openstack/ironic stable/2026.1: Security: Fix sensitive properties returned on volume targets https://review.opendev.org/c/openstack/ironic/+/992321 | 17:23 |
| opendevreview | Julia Kreger proposed openstack/ironic bugfix/34.0: Security: Fix sensitive properties returned on volume targets https://review.opendev.org/c/openstack/ironic/+/992322 | 17:31 |
| opendevreview | Julia Kreger proposed openstack/ironic bugfix/33.0: Security: Fix sensitive properties returned on volume targets https://review.opendev.org/c/openstack/ironic/+/992323 | 17:32 |
| opendevreview | Julia Kreger proposed openstack/ironic stable/2025.2: Security: Fix sensitive properties returned on volume targets https://review.opendev.org/c/openstack/ironic/+/992325 | 17:35 |
| opendevreview | Julia Kreger proposed openstack/ironic stable/2025.1: Security: Fix sensitive properties returned on volume targets https://review.opendev.org/c/openstack/ironic/+/992326 | 17:36 |
| opendevreview | Julia Kreger proposed openstack/ironic unmaintained/2024.1: Security: Fix sensitive properties returned on volume targets https://review.opendev.org/c/openstack/ironic/+/992333 | 17:52 |
| opendevreview | Julia Kreger proposed openstack/ironic unmaintained/2023.1: Security: Fix sensitive properties returned on volume targets https://review.opendev.org/c/openstack/ironic/+/992335 | 18:18 |
| opendevreview | Merged openstack/ironic master: Fix kernel parameter parsing for quoted values and whitespace https://review.opendev.org/c/openstack/ironic/+/992025 | 18:20 |
| TheJulia | Random off the wall question, would it make sense to have a sort of mid-cycle long discussion call to enable mid-cycle alignment for contributors? | 18:33 |
| JayF | I don't hate the idea; I'd hit the list with it though so we catch contributors who are usually sleeping now | 18:37 |
| TheJulia | true | 18:39 |
| opendevreview | Clif Houck proposed openstack/ironic stable/2026.1: Fix kernel parameter parsing for quoted values and whitespace https://review.opendev.org/c/openstack/ironic/+/992340 | 18:54 |
| opendevreview | Merged openstack/ironic-python-agent bugfix/11.6: Update .gitreview for bugfix/11.6 https://review.opendev.org/c/openstack/ironic-python-agent/+/992207 | 19:09 |
| opendevreview | Clif Houck proposed openstack/ironic stable/2025.2: Fix kernel parameter parsing for quoted values and whitespace https://review.opendev.org/c/openstack/ironic/+/992341 | 19:24 |
| opendevreview | Clif Houck proposed openstack/ironic stable/2025.1: Fix kernel parameter parsing for quoted values and whitespace https://review.opendev.org/c/openstack/ironic/+/992345 | 19:41 |
| opendevreview | Jacob Anders proposed openstack/ironic-specs master: Add spec for multi-component batched firmware updates https://review.opendev.org/c/openstack/ironic-specs/+/992208 | 21:16 |
| janders | JayF TheJulia updated https://review.opendev.org/c/openstack/ironic-specs/+/992208 (firmware updates improvement spec) to link to the RFE/bug - thanks for pointing that out | 21:18 |
| janders | (had the reference in the spec itself but forgot to put it on the change also) | 21:18 |
| opendevreview | Clif Houck proposed openstack/ironic bugfix/34.0: Fix kernel parameter parsing for quoted values and whitespace https://review.opendev.org/c/openstack/ironic/+/992367 | 21:40 |
| sean-k-mooney | TheJulia: midcycels can be prodcutive, adopting a 6-10 week release cycle like the kernel would be even better IMO | 21:40 |
| sean-k-mooney | but ya i have codnierd doign one in the past, it been a long time since i attened a midcycle for any of the project im in vovled with regually but my memory was they were useful | 21:42 |
| TheJulia | Ironic already does that, fwiw. Anyhow, I was more proposing the idea of a long video call to sync folks up because of just how insane this cycle is turning out | 21:42 |
| sean-k-mooney | ya i was thinkign fo proposign one for the agentic workflows repo because we didnt really have time at the ptg and we have all been too busy to get back to that topic | 21:43 |
| TheJulia | Anyway, back to exercising becuase they won't let me find out if I have to report for jury duty tomorrow for a few more hours *cries* | 21:43 |
| sean-k-mooney | so i would like to do something this cycle there but i feel like we need to get concenous on it first | 21:44 |
| TheJulia | Yeah, that is the downside of the PTG Model... without the human interaction, you really have to keep hammering items forward | 21:44 |
| TheJulia | because your also building mindshare/critical mass | 21:44 |
| TheJulia | and that is even harder without the ability to hallway track and share context | 21:44 |
| sean-k-mooney | yep the lack of the hallway/evening track too do that is a downside | 21:44 |
| opendevreview | Clif Houck proposed openstack/ironic bugfix/33.0: Fix kernel parameter parsing for quoted values and whitespace https://review.opendev.org/c/openstack/ironic/+/992368 | 21:46 |
| TheJulia | It is interesting, because at the Summit last year, I was able to have some hallway track discussions which then got carried forward into standards bodies discussions to generally just get things to a better place overall. | 21:46 |
| TheJulia | Anyway, exercise, the clock is still running | 21:46 |
| sean-k-mooney | o/ | 21:47 |
| david | Hello, I am new to Ironic but am very happy with what I have working so far with it. One of the final steps for us is getting it working with MDADM raid for the / volume. We normally use LVM here to allow resizing of partitions over the life of the instance. From what I'm seeing in the docs this isn't currently supported. Is the correct approach here to work on writing a implementation using a custom HardwareManager and building | 22:11 |
| david | work on this project so far, it's far more advanced that any of the other bare metal management tools I've looked at over the years with Foreman/Maas. | 22:12 |
| TheJulia | Hi david | 22:14 |
| TheJulia | david: I guess it sort of depends on where your coming from and how much ?pain? (is pain the right word? I'm not really sure) your willing to deal with. Ultimately RAIDing on deploy is a difficult topic made worse by ecosystem constraints outside of Ironic. For example, default bootloader binaries lacking raid support out of the box. | 22:16 |
| TheJulia | Which is sort of why RAID support is in a documented limbo of sorts, people obviously see it as a solution, but... It'ts not always the best solution and in the end its a bit harder to support project wise. Granted, there are RAID operators out there. CERN specificially does do raid boot and root volumes and uploads patches from time to time. | 22:17 |
| david | TheJulia: Thanks. Currently i'm building the image with packer and kickstart files to setup the base system, packages, install mdadm, cloud init etc. Our typical deployment is just mirror two boot devices with 3 partitions on each disk. (part1: for ESP (no raid, seperate mount points and a cron rsync job to keep the ESP in sync.) part2: 2GB /boot raid1 mdadm, part3: raid1 mdadm that houses pv.0 where we make the lvm partitions fo | 22:24 |
| TheJulia | fair enough, that should basically work | 22:26 |
| TheJulia | https://docs.openstack.org/ironic/latest/admin/raid.html#software-raid is the latest with a nice warning that I was alluding to | 22:28 |
| TheJulia | Thats a little different, fwiw, but if there are mild changes which you could use to the RAID support in the agent, we could likely accept them into the codebase without too much hassle. If you want to match what you've already been doing exactly, I'd recommend a hardware manager, but I've not tried anything like that as it relates to raid. | 22:30 |
| david | Yup thanks, I was looking at thouse docs and came across the bullet point of "Images utilizing LVM for their root filesystem are not supported. Patches are welcome to explicitly support such functionality." Which led me to wonder why lvm root works for a non mdadm build (I have it working in my testing just fine.) but apparently doesn't when using the buildin raid support in ironic. | 22:31 |
| david | Thanks for pointing me in the right direction. I'll start looking at the source code and see where I land at for writing something. There's a few other features I'd love to see such as defining which disks to mdadm based on disk vendor ID/serial on a per-node basis. (We have a lot of nodes where there's 2 drives used for boot and lots of other drives used for ceph osds for example etc.) | 22:36 |
| TheJulia | Oh, oh! Yeah, LVM based… I haven’t tried. I wonder if CERN has tried. | 22:36 |
| david | Yeah for us we use a mdadm to build a /boot. then we build another mdadm mirror out of part3 and create a LVM PV on that which we then create partitions for /var /var/log and /. | 22:37 |
| TheJulia | Yeah, disk selection wise, check the ironic-python-agent source. I think the software raid code is much simpler but with a device ignore model | 22:37 |
Generated by irclog2html.py 4.1.0 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!